…We need to encourage women not to be afraid to fail. Many women hold themselves back because they’re not sure they can do the job. But approached the right way, failure is an essential bridge to success, so women have to be willing to push through the discomfort and be bold and find a way to succeed.
As part of my series about the women leading the Artificial Intelligence industry, I had the pleasure of interviewing Dr. Judith Bishop.
Dr. Judith Bishop is Senior Director of AI Specialists at Appen. She heads a team of linguists and language experts creating AI training data to support the development of voice, text and multimodal AI interfaces that work for diverse users in real world environments.
Thank you so much for doing this with us! Can you share with us the ‘backstory” of how you decided to pursue this career path?
I’m proud to say that Dr. Julie Vonwiller, the co-founder of Appen, headhunted me 16 years ago! She was looking for someone skilled in prosodic annotation, and I was recommended to her. I started out as a linguist supervising prosodic annotations for Toshiba Research Labs across 6 different languages, became a project manager for a couple of years, then a senior linguist supervising the more high-profile and complex projects. In 2010, I started managing a group of 6 linguists and now direct a group of about 100 talented linguists and language experts who are helping to reshape how linguistics contributes to the evolution of AI.
What lessons can others learn from your story?
I feel I’ve personally learned that patience and dedication to a chosen career will pay off over time. I always thought that my linguistic study could lead to industry applications — but nothing could have prepared me for how rich and deep the implications of linguistics are for current developments in AI. By staying on this career path, I’ve been privileged to witness the transformation of language technology into something that will be absolutely central to how we live and interact with machine intelligence in future. There are also such benefits to being part of the memory of an organization. In the case of Appen, it feels like familial memory: I remember individuals and their contributions, as well as where all kinds of knowledge can be found and how it was developed by Appen. There are many of us at Appen with this depth of memory: it’s that kind of company. As well as being a bonding fabric, this continuity of experience translates into some exceptionally well-developed processes for ensuring the quality of our work.
Can you tell our readers about the most interesting projects you are working on now?
There are no uninteresting projects in AI, and while I can’t give specifics, we’ve started to move into multimodal data creation and product evaluation projects that require a new focus on time and timing — for example, how the visual output of the AI engine, such as an avatar face or face and body, aligns in time with the synthesized speech. Human beings are super keyed into timing. It’s usually at a subliminal level, but as you know from watching poorly dubbed movies, any slight misalignment between channels can really make the experience challenging.
None of us are able to achieve success without some help along the way. Is there a particular person who you are grateful towards who helped get you to where you are? Can you share a story about that?
Yes. Julie Vonwiller was the one who first encouraged me to set my sights high. She did this both in words and in action: each time I returned from maternity leave I was given greater seniority and opportunities to grow. This happened even though I returned to work part-time for a period after each leave. It gave me such confidence that my knowledge and commitment was valued by Appen and its management. It also felt like a recognition of what many parents know: that becoming a parent focuses your energies in a way that benefits your career. Without Julie’s support, I doubt I would be who I am today. Another leader I am grateful to is Bill Pulver, a previous Appen CEO. When Bill was new in the CEO position, I came to him with a problem: there was a leadership gap in my team that was causing problems. He listened then asked me if I thought I could fix it. I said I could; and that’s how I began the management phase of my career.
What are the 5 things that most excite you about the AI industry? Why?
What excites me most about AI is that it’s aimed both at mimicking and exceeding human capabilities of reasoning, perception and communication. The knowledge we’ll gain about ourselves through developing AI — how we interact with each other and with the world — will be so deep. Linguists will be critical to this process because we’ll have to unpack how we communicate in order to help machines learn what we do and how we understand the world.
Another exciting area of development is that, with COVID-19, contactless services will continue to be important to all types of service providers. As we look to convert any situation where you currently have to walk into a room and talk with someone into a contactless service, we’ll continue to look at ways that AI can power chatbots, virtual assistants and the like. In healthcare, for example, there’s tremendous interest in developing AI-powered healthcare assistants that can perform triage to provide more options that eliminate the need for patients who are feeling ill to travel to see a doctor. This type of benefit will be replicated across many industries.
What are the 5 things that concern you about the AI industry? Why?
I would bucket my concerns into two main categories. First is the fact that policy and regulation are lagging far behind AI development. There’s a very real possibility that AI will leave people vulnerable in ways they’ve never been before. AI-powered insights into how we communicate and what we’re expressing and the emotions we’re expressing it with — things most people will want to keep hidden — will be stored in systems and devices. This can create opportunities for abuse.
It’s also a possibility that we could fail to ensure that AI technology is fair in terms of diversity and inclusivity. Whether intentionally or unintentionally, if we use biased training data for AI systems, we could end up replicating existing human bias in powerful AI projects that treat people unfairly on a huge scale. Systemic bias in automated systems for mortgage and credit applications, insurance rates, and even hiring come immediately to mind.
As you know, there is an ongoing debate between prominent scientists, (personified as a debate between Elon Musk and Mark Zuckerberg,) about whether advanced AI has the future potential to pose a danger to humanity. What is your position about this?
Any new technology can be used to do harm. As I said, with AI, biased training data can be a source of danger to individuals and groups. AI also requires the reliance on millions of people from around the world to annotate the AI training data, raising the issue of how to treat these people fairly.
What can be done to prevent such concerns from materializing? And what can be done to assure the public that there is nothing to be concerned about?
It is up to governments, scientists and especially every enterprise developing AI projects to commit to responsible AI, principles advanced by Appen and the World Economic Forum (WEF) that set out standards and best practices for the training of data. Responsible AI will improve data quality and the efficiency and transparency of AI development, while eliminating bias. Responsible AI also includes paying annotators fair wages and adhering to labor wellness guidelines and standards. It’s really to everyone’s benefit to be transparent about the potential downsides of AI and then work together to avoid them.
How have you used your success to bring goodness to the world? Can you share a story?
Recently, my team was invited to be part of the TICO-19 COVID resources translation effort run by Translators without Borders and a consortium of major technology companies. The team performed translation QA on more than 950,000 words across 38 mainly under-resourced languages, many of them from Africa, such as Dari, Dinka, Hausa, Luganda, Pashto, Zulu, etc. I was able to use my position to advocate for Appen supporting this pro-bono effort. It’s been a wonderful experience and platform for demonstrating our exceptional expertise in translation and low-resource languages.
As you know, there are not that many women in your industry. Can you share 3 things that you would you advise to other women in the AI space to thrive?
First, we need more executive support for leadership programs for women in AI. Execs should always assume that a talented woman is ready for a leadership role — or at least is ready to be trained in that direction — and they need to make these women more visible both in the organization and externally.
Second, we need to encourage women not to be afraid to fail. Many women hold themselves back because they’re not sure they can do the job. But approached the right way, failure is an essential bridge to success, so women have to be willing to push through the discomfort and be bold and find a way to succeed.
Third, women need to give themselves time to develop the skills and do the research to be able to work in this industry and be leaders in this industry. This is possibly the hardest one, especially given the pace of development of AI, since women have so many different roles in life. But we need to set aside those few hours now and then for ourselves.
Can you advise what is needed to engage more women into the AI industry?
More women in visible positions in AI leadership will have the single biggest impact. It’s a bit of a chicken-and-egg problem, but there are so many talented women who could be AI leaders. That’s where executive support for women’s leadership development comes in.
What is your favorite “Life Lesson Quote”? Can you share a story of how that had relevance to your own life?
My favorite life lesson quote is actually a proverb: ‘A stitch in time saves nine.’ Like most proverbs, it distils a lot of human experience into a few words. The importance of taking the right action at the right time resonates for me on so many levels. At work, it’s a constant reminder that achieving quality outcomes means spending time upfront planning and preparing — and automating anything that’s going to slow us down. Many years ago, we were spending a lot of time correcting words that were spelled in different ways in transcription projects. There was no easy solution, but we put together a multi-pronged approach to find all the related spellings and link them to a single key spelling — automatically wherever possible. Our approach worked for any language. A single key spelling for each word turned out to be a mission-critical requirement for one of our largest, most complex and strategically important transcription programs in the following years. We would have failed to deliver if we hadn’t put in the effort and created that solution at the right time.
You are a person of great influence. If you could start a movement that would bring the most amount of good to the most amount of people, what would that be? You never know what your idea can trigger. 🙂
My movement would aim to ensure that AI and language technology developments actively support the future of the 7,000 languages that currently exist in the world. Fortunately, this movement is already underway through the positive action of companies such as Microsoft, Amazon, Google and Facebook, but it will need concerted focus and long-term philanthropic commitment from companies and governments around the world. There is so much work to be done, and soon, before the precious human heritage of language diversity is lost to future generations. According to the UN, roughly 97% of the world’s population speaks about 4% of those 7,000 languages. What that means is that for most of the world’s languages, the development of AI and language technology isn’t supported by a commercial imperative. But even that 4% is 280 languages, and only a fraction of those languages is currently well-served by AI core technologies such as automated speech recognition, speech synthesis, natural language understanding and natural language generation. That’s an enormous number of people who can’t access these AI technologies through their native tongue(s). Just imagine the positive impact on speakers of these languages if they could speak or otherwise interact with AI devices in the language of their choosing.
How can our readers follow you on social media?
Your readers are very welcome to follow me on LinkedIn or through Appen’s social media posts.
This was very inspiring. Thank you so much for joining us!